| Mobile Web

Search results for Reinforcement learning

AI & Enterprise

Rethinking AI performance scaling beyond parameter count

Jie Tang, a Tsinghua University professor and founder of AI startup Z.ai, said AI model performance is not determined by parameter count alone. He said data volume, how computing resources are used, and operating conditions also matter. Tang pointed to shifts from Jared Kaplan’s 2020 scaling results to DeepMind’s 2022 Chinchilla law and argued that inference-driven lifetime costs can favor smaller models trained longer. He also described how MoE architectures change scaling trade-offs.

AI & Enterprise

AI data race moves beyond chatbots to learning how humans work

The race to secure AI training data is shifting from improving chatbot answers to teaching systems how people actually work, Business Insider reported. Large language models have advanced using internet text, books and human ratings, but that approach has limits for building AI agents that handle complex work over hours or days. Companies are turning to reinforcement learning environments that simulate workplaces and reward task completion. Google, Tesla, Meta, Scale AI and Uber are cited as pursuing such efforts.

AI & Enterprise

Musk counterattack: Grok 4.6 ahead of GPT in coding and terminal performance

SpaceXAI unveiled its new AI model Grok 4.6. Foreign media reports said it matched OpenAI\'s GPT-5.6 Sol Max on a composite performance score and focuses on agent use and conversational and visual outputs. In Artificial Analysis\' Intelligence Index, Grok 4.6 scored 61, behind Anthropic\'s Fable5 Max by 1 point. Grok 4.6 led in software engineering and terminal benchmarks and will be available via multiple channels with pricing per million tokens.